1. Field of the Invention
The present invention relates to power semiconductor device and, more particularly, to an insulated gate static induction thyristor with a split gate type shorted cathode structure wherein a first gate (a shielding gate) is shorted to a cathode region formed in a second gate and an insulated gate formed over the second gate is used as a control gate electrode to enhance its isolation from a cathode short-circuit gate, thereby increasing the maximum controllable current/voltage durability of the device and permitting its high-speed switching.
2. Description of the Prior Art
It is well-known in the art that the above-mentioned maximum controllable current/voltage durability of a conventional thyristor, GTO, or similar device could be enhanced by the introduction thereinto of a shorted cathode structure. As for a static induction thyristor with a shorted cathode structure, there has been proposed a device structure for a planar gate type SI thyristor and it has been reported that speedup of device operation could be implemented without shortening the carrier life time in the device by the use of a double emitter-shorted structure combined with a shorted anode structure. FIG. 25 is a sectional view schematically showing the internal construction of an SI thyristor with a double emitter-shorted structure disclosed in "Switching Characteristics of an SI Thyristor with a Shorter Structure" EDD-90-59, SPC-90-58 presented in Joint Meeting on Electron Devices and Semiconductor Power Conversion, IEE of Japan.
In FIG. 25 reference numeral 1 denotes a p.sup.+ -type anode region, 2 an n.sup.+ -type cathode region, 3 a p.sup.+ -type gate region, 4 a cathode short-circuit region, 5 a high-resistivity region, 6 an n.sup.+ -type static induction short-circuit region, 7 an anode electrode, 8 a chathode elctrode, 9 a gate electrode, 10 a cathode short-circuit electrode, and 11 an oxide film.
In the conventional shorted cathode structure, as shown in FIG. 25, the cathode short-circuit region 4 is provided separately of the p.sup.+ -type gate region 3 so that holes to be absorbed into the p.sup.+ -type gate region 3 are partly absrobed by the cathode short-circuit region 4 to essentially reduce the hole density near the cathode side. Since the cathode short-circuit region 4 is spaced a predetermined distance apart from the p.sup.+ -type gate region 3, however, the areas of the cathode region 2 and the p.sup.+ -type gate region 3 are smaller the in a thyristor structure employing a common gate, and hence the area efficiency decreases accordingly. The decrease in the area of the cathode region 2 will cause a corresponding decrease in the current capacity of the thyristor as well. Hence, this prior art example is so low in area efficiency that it is not suited to providing a large current by a multi-channel structure of the device.
In Japanese Patents NO. 1588399 entitled "Static Induction Thyristor" and No. 1456781 entitled "Double Gate Type Static Induction Thyristor" there are disclosed split gate type SI thyristors of the type wherein a gate region surrounding a channel is split into a plurality of gates and one of them is used as a driving gate to sufficiently reduce the electrostatic capacitance viewed from the gate, permitting high-speed switching of large current. Yet, these patents disclose only enhancement of the function of a control gate by dividing the gate but does not ever disclose the shorted cathode structure which positively enhances the function of a non-control gate to improve the maximum controllable current/voltage durability.
A split gate type static induction transistor is also disclosed in Japanese Patent No. 1302727 entitled "Static Induction Transistor and Semiconductor Integrated Circuit," for instance; but the invention disclosed in the patent is directed to transistors and is applied to integrated circuits that operate at low voltages. Although the invention permits improvement of the function of the control gate, nothing is disclosed about measures for improving the function of the non-control gate. Moreover, the patent does not ever disclose anything about a shorted source structure that corresponds to the shorted cathode structure according to the present invention. The reason for this is that the non-control gate is not so important because of the split gate structure of the transistor.
On the other hand, a semiconductor imaging device that utilizes the split gate type transistor as one pixel (a picture element) has also been proposed and is disclosed in Japanese Pat. Publication Gazette No. 37028/89. In this imaging device, the gate is split into a control gate and a shielding gate; the shielding gate is used as a region for device isolation and shields against irradiation by light, and the control gate is used as an ordinary gate and is covered with a capacitor that is used to store optical information.
In the above-mentioned Publication Gazette there are proposed, with a view to separating the functions of the two gate regions, a structure in which the distance between the shielding gate region and the source (drain) region formed as a main electrode in the wafer surface is short and a structure in which the shielding gate region is formed deep.
However, such structures for isolating the functions of split gates have not been proposed for semiconductor devices that operate under high withstand voltage, large current and high intensity electric field conditions, such as thyristors. In the case of the avobe-mentioned imaging device or split gate type static induction transistor, the voltage that is applied to the main electrode is as low as 5 volts or so and conducting carriers are electrons (in the case of an n-chennel device). The effect of storage of holes that are carriers injected from an pn junction control gate is less than in the case of using an common gate. In the split gate type static induction thyristor, however, conducting carriers are both electrons and holes and there are holes injected from the anode region into the cathode region as well as holes injected from the control gate that have an influence on the moinority carrier storage effect in the vicinity of the cathode side, in particular. In conventional split gate type static induction thyristors there have not been taken any measures for solving a problem such as how to control the hole current from the anode region by the shielding gate to increase the maximum controllable current/voltage durability.
The use of the split gate structure reduces the quantity of holes that are injected from the control gate, but the holes that are injected into the cathode region are mostly those from the anode region.
In connection with split gate type static induction thyristors, it is well-known that splitting of the gate reduces the input capacitance and increases the mutual conductance Gm at the time of electron injection from the cathode region, a decrease in the RC time constant by the reduced input capacitance providing high-speed turn-OFF performance; but no structure has been proposed which improves the turn-OFF performance by separating the functions of the split gates to provide for increased maximum controllable current/voltage durability. Nor has there been proposed any structure wherein the holes injected from the anode side are positively flowed into the shielding gate region to lighten the burden on the control gate and prevent the turn-OFF characteristic from being suppressed by the minority carrier storage effect near the cathode side.
To solve the above-mentioned defects of the prior art, the inventors of this application have previously proposed, in Japanese Pat. Appln. No.289244/92, a static induction thyristor with a shorted cathode structure wherein first and second gates are both formed as pn junction gates. The static induction thyristor disclosed in the above-said prior application is a static induction thyristor with a split gate type shorted cathode structure which lessens the minority carrier (hole) storage effect near the cathode side to attain high-speed turn-OFF performance, increase the maximum controllable current/voltage durability and permit high-speed switching.
More specifically, the channel is surrounded by two split gates, one used as a control gate and the other as a cathode short-circuit gate electrically shorted to the cathode. As compared with conventional structures, this structure provides a high channel integration density and consequently a high area efficiency, lessens the minority carrier storage effect to increase the switching speed, and increases the maximum controllable current/voltage durability by the cathode short-circuit effect.
The static induction thyristor with a split gate type shorted cathode structure, disclosed in the above-mentioned prior application, has an anode region, a cathode region and a control region formed with a high-resistivity region interposed therebetween. The control region includes first and second gate regions separated from each other. A shield gate electrode formed in contact with the first gate region and a cathode electrode formed in contact the cathode region are electrically shorted to form a shorted cathode structure. A current flow between the cathode region and the anode region is controlled by a voltage that is applied to a control gate electrode formed in contact with the second gate region. In the high-resistivity layer adjoining the cathode region there are formed a first depletion layer by the diffusion potential between the first gate region and the high-resistivity layer and a second depletion layer by the diffusion potential between the second gate region and the high-resistivity layer. At the same time, a potential barrier, which is conrollable with a static induction effect by the voltage of the control gate electrode that is applied to the second gate region, is formed in the high-resistivity layer near the boundary between the first and second depletion layers. Holes injected from the anode region partly flow through the first gate region and into the cathode electrode shorted to the shield gate electrode.
The first gate region has an impurity concentration higher than that of the second gate region.
The distance between the first gate region and the cathode region is selected to be shorter than the distance between the second gate electrode and the cathode region; namely, the cathode region is formed closer to the first gate region than to the second gate region.
The first gate region includes a medium or low impurity concentration diffused region of the same conductivity type as that of the first gate region, formed deeper than the second gate region, and a high impurity concentration diffused region formed in the medium or low impurity concentration region.
The first gate region is formed deeper and wider than the second gate region.
The cathode region is separated from the first and second gate regions by a medium or low impurity concentration region of the same conductivity type as that of the cathode region and formed around the cathode region.
One or both of the first and second gates have a buried gate structure.
Alternatively, the first and second gate regions both have the buried gate structure, and the first and second gate region are adjacent each other.
Alternatively, one or both of the first and second gate regions have a recessed gate structure.
Alternatively, the cathode region, the anode region and the first and second gate regions are all formed in the vicinity of the same main surface of the wafer.
FIG. 26 is a schematic cross-sectional view for explaining the principle of operation of the static induction thyristor disclosed in the afore-mentioned Japanese prior application. FIG. 27 is a schematic equivalent circuit representation of the static induction thyristor shown in FIG. 26. As will be seen from FIG. 27, the operation of the static induction thyristor of FIG. 26 can be regarded as parallel operations of two thyristors since it has the split gate structure. In FIG. 26, reference numeral 1 denotes an anode region, 2 a cathode region, 31 a first gate region that is called a cathode short-circuit gate or shielding gate, 32 a second gate region that is called a control gate, 5 a high-resistivity layer, 7 an anode electrode, 8 a cathode electrode, 9 a gate electrode, 10 a cathode short-circuit electrode, and 11 an oxide (SiO.sub.2) film. The gate electrode 9 is a control gate electrode, and the cathode short-circuit electrode 10 is the electrode of the cathode short-circuit gate (or shielding gate) and is essentially shorted to the cathode electrode 8.
Of course, it does not matter, theoretically, whether the gate regions 31 and 32 are pn junction gates, MIS (MOS) gates, Schottky gates, or hereto junction gates. They need only to have a gate structure that permits control of current between the cathode 2 and the anode 1 by the static induction effect. In FIG. 26, reference character W.sub.1 indicates the width of a depletion layer spreading around the first gate 31 and W.sub.2 the width of a depletion layer around the second gate 32. Reference character G* indicates what is called an intrinsic gate point, which corresponds to the top of a static induction barrier height. The potential barrier near the point G* acts as a barrier against holes present in the first and second gates as well as electrons in the cathode, For instance, when the thyristor is in the OFF state, a potential barrier of a sufficient height is formed near the point G* against the electrons in the cathode, while at the same time it also serves as a potential barrier against the holes in the first gate 31 and the second gate 32. Consequently, when the thyristor is in the OFF state, no electron current conducts between the anode and the cathode, and no hole current conducts between the first and second gates 31 and 32 either.
As the potential harrier height near the point G* is decreased by the application of a positive voltage to the control gate electrode 9 of the second gate 32, an electron injection from the cathode 2 begins. Holes are also injected from the second gate 32, but the quantity of holes injected is far smaller than the quantity of holes that are injected from the anode 1 afterward. Besides, the quantity of holes that are injected from each split gate is also smaller than in the case of employing a common gate, which is an advantage of the split gate structure, the carrier storage effect of the gate is small. Furthermore, since the potential barrier against the holes is present near the point G*, the hole current from the second gate 32 to the first gate 31 is mainly a displacement current accompanying capacitive coupling, and an essential conducting current is very small. As the electrons injected from the cathode 2 are stored near the interface between the anode region 1 and the high-resistivity layer 5 and the height of the potential barrier against the holes in the anode region 1 decreases accordingly, injection of holes from the anode region 1 starts. The hole current from the anode region 1 mostly flows into the first gate 31 electrically shorted to the cathode region 2, the remaining hole current flowing into the second gate 32. The rate at which the hole current flows into the first and second gates 31 and 32 is dependent of the configuration of the potential distribution by the relative potential difference between the first and second gates 31 and 32, their area ratio or their shapes such as geometrical depths. When the second gate 32 is held at a positive potential relative to the first gate 31, it is expected that the quantity of holes flowing into the first gate 31 will be essentially larger than to the second gate 32. The reason for this is that the first gate 31 is lower in potential than the second gate 32 and hence is essentially in a state in which it readily stores the holes. However, the second gate 32 that serves as a control gate is small in electrostatic capacitance, and hence is readily charged by a relatively small hole current. This is another advantage of the split gate structure. In consequence, the potential at the point G* further decreases, causing further injection of electrons and further supply of holes from the anode electrode 7 and the anode region 1. By this, the thyristor is put into a latch-up state, in which the potential at the point G* decreases, permitting the formation of a channel for electrons between the anode region 1 and the cathode region 2. On the other hand, the potential barrier against holes increases, resulting in a high potential barrier being formed between the first and second gate regions 31 and 32.
That is, substantially no hole current flows between the first and second gate regions 31 and 32. Hence, when thyristor is in the ON state, electrons from the cathode region 2 flow into the anode region 1 and thence to anode electrode 7, whereas holes from the anode region 2 flow into the first gate 31 and the cathode region 2, respectively.
Next, a description will be given of the turn-OFF operation of the thyristor. When a negative electrode is applied to the gate electrode 9, the width W.sub.2 of the depletion layer spreading out from the second gate 32 into the high-resistivity layer 5 increases, the potential barrier height near the point G* increasing. As a result, the hole current that has flowed into the cathode region 2 and the first gate retion 31 from the anode region 1 so far partly flows into the negatively biased second gate region 32 and thence to the control gate electrode 9. Yet, the quantity of the hole current that flows into the second gate region 32, by-passing the first gate region 31, is far smaller than the total amount of hole current flowing into the first gate (the short-circuit gate) region 31, rather, the injection of electrons from the cathode region 2 stops, since the potential barrier height bear the point G* is instantaneously increased by the negative bias voltage applied to the second gate region 32. In this state the hole current from the anode region 1 mostly flows into the first gate (or short-circuit gate) region 1 but the quantity of this current gradually decreases. The hole current having flowed into the cathode region 2 so far flows into the negatively biased second gate (or control gate) region 32. Thus, the entire hole current is shared between the first and second gate regions 31 and 32 because of the split gate structure, and hence the amount of hole current that the control gate (or second gate) region 32 needs to control may be far smaller than in the case where a common gate structure is used. Moreover, the use of the split gate structure reduces the quantity of holes that are injected from the respective gate, and hence the minority carrier (holes) effect is lessened accordingly.
To stop the hole injection from the anode region 1 after the electron injection from the cathode region 2 has also been stopped by the restoration of the poitential barrier at the point G* to its original height by negative biasing of the second gate region 32, it is necessary to extinguish the electrons stored near the anode region 1 through structural or lifetime control and through use of a structure which stops the hole injection from the anode region 1 (an SI anode short, double gate structure, for instance) or by effecting hole lifetime control.
Since the first gate region 31 is always electrically shorted to the cathode region 2, holes near the first gate region 31 are readily absorbed thereinto. Hence, the hole storage effect near the cathode region 2 and near the first gate region 31 is insignificant. Moreover, since there is no potential difference between the first gate region 31 and the cathode region 2, the width of the depletion layer spreading between them is substantially constant, the capacitance between them undergoing little change. Hence, the capacitance between the second gate region 32 and the cathode region 2 more greatly contributes to the switching operation of the thyristor. On the other hand, the width N, of the depletion layer spreading from the first gate region 31 toward the anode region 1 undergoes a substatial change with the voltage condition in the anode region 1, and this causes a substantial change in the capacitance between the first gate region 31 and the anode region 1; it is preferable that the influence of such a large capacitance change be prevented from affecting the cathode side.
In the static induction thyristor with a shorted cathode structure, disclosed in the afore-mentioned prior Japanese patent application, the anode-to-cathode current that is controlled by the control gate (or second gate region) 32 can essentially be increased by the effect of the first gate (or short-circuit gate) region 31 which serves as a hole absorbing region, and hence the maximum controllable current durability can be expected to increase. Furthermore, the amount of hole injection from the control gate 32 is small and the number of carriers (holes) stored near the cathode region 2 during the ON state of the thyristor is also made virtually smaller by the shorted gate structure than in the case where the common gate structure is utilized. Hence, the amount of holes to be absorbed by the control gate 32 during the turn-OFF of the thyristor may be so small that the turn-OFF state could be improved. Besides, it is also expected as another advantage of the split gate structure that the turn-ON switching performance is improved by the reduction of the gate input capacitance.
With the above-described split-gate, shorted-cathode structure, it is possible to implement a static induction thyristor which is high in the area efficiency because of the high channel integration density, high in the switching speed because of the lessened minority carrier storage effect and high in the maximum controllable current/voltage durability because of the use of the shorted cathode structure. The split gate structure can be formed as a planar, buried, recessed or double gate structure, and it is applicable to medium, small and large power semiconductor devices and high withstand voltage integrated circuits as well.
In the case of using the structure wherein the first and second gate regions 31 and 32 are both formed by pn junction gates, however, it is necessary, for separating their functions, that they be formed with different impurity densities or in different sizes, for instance. On the other hand, the first gate region 31, which is formed as a pn junction gate for absorbing holes injected from the anode region 1, is indispensable to the thyristor, but the second gate region 32 as the control gate need not always be a pn junction gate, because the function of the second gate 32 as the control gate is to control the electron injection from the cathode region 2 through use of a potential barrier. Where the control gate is formed by a pn junction gate, the hole injection from the control gate incurs the storage of extra minority carriers. In addition, it is difficult to completely isolate the pn junction of the control gate from the pn junction forming the first gate region 31. In other words, the potential of the second gate region 32 is affected by the potential of the first gate region 31, and hence it is hard to attain independent controllability of the control gate 32.
In view of the above, the inventions of this application propose a novel structure that has the cathode region 2 formed virtually in the second gate region 32 and controls the injection of electrons from the cathode region 2 by an insulated gate. That is the second gate region 32 is not used intact as a control gate but instead an insulated gate formed over the second gate region 32 with an insulating film interposed therebetween is used as the control gate. With this structure, the injection of minority carriers from the control gate electrode is essentially suppressed, because of a MOS insulating layer interposed between the second gate region 32 and the control gate electrode. An electron channel region is an inversion layer in the MOS gate interface or deep channel region and its controllability is excellent because the MOS gate electrode acts as a control electrode. The inversion layer in the MOS gate interface is formed in the surface layer of the second gate region 32. Hence, it can be said that an equivalent MOSFET is formed or defined by the n-type cathode 2, the p-type second gate region 32 and the n.sup.- type high-resistivity region 5. This equivalent MOSFET may also be formed at the side opposite from the first gate region 31. The high-resistivity region 5 between the first and second gate regions 31 and 32 is virtually depleted, and in this high-resistivity region 5 there may be formed, by voltage applied to the MOS gate electrode, a potential barrier controllable by the static induction effect. In this instance, the potential of the second gate region 32 will be controlled static-inductively by capacitive coupling on the basis of the potential of the first gate region. However, the MOS gate electrode and the first gate region 31 are isolated from each other substantially completely by the insulating layer formed all over the second gate region 32.